Abstrato

Abnormal walk identification for systems using gait patterns

Sithi Shameem Fathima SMH, Wahida Banu RSD

Gait based person recognition is an emerging technique in this recent era. Most of the biometrics system requires high resolution and proximity sensors. Gait based identification is used for identification, authentication and gender calculation. Due to various reasons, 5 to 15% of persons are not having normal walk pattern because of drug abuse or habitual and medical rehabilitation. So it is mandatory to classify normal or abnormal walk. Steppage gait, circumduction gait, Charlie Chaplin gait and antalgic gait are taken into consideration for this data base. In this modern time by using computer vision based techniques, normal and abnormal walk pattern can be identified. This paper considers gait cycle of a person of abnormal walk and his height and width are taken as feature vectors. K-Nearest Neighbour (KNN), Convolutional Neural Networks (CNN) and Artificial Neural Network (ANN) Classifiers are used to classify normal and abnormal gait patterns. The Artificial Neural Network (ANN) classifier outperforms well than K-Nearest Neighbour (KNN) and Convolutional Neural Networks (CNN) classifiers and it produced good recognition rate using this feature vectors.